Patient-Specific Imaging and Dosimetric Errors in PET/CT-Guided Radiotherapy of Lung Cancer
S R Bowen*, M J Nyflot, J Do, J Meyer, P E Kinahan, G A Sandison, University of Washington, Seattle, WASU-D-141-6 Sunday 2:05PM - 3:00PM Room: 141
Purpose: PET/CT guided-radiotherapy of lung cancer requires estimation and mitigation of errors due to respiratory motion. A patient-specific workflow was developed to measure uncertainties in imaging, treatment planning, and radiation delivery with respiratory motion phantoms and dosimeters.
Methods: A torso phantom with inserts mimicking normal lung tissue and lung lesion was filled with [¹⁸F]FDG. The lung lesion insert was driven by patient-specific respiratory patterns or kept stationary. PET/CT images were acquired under static, motion (3D), and respiratory phase-gated (4D) conditions and reconstructed with OSEM (2 iterations, 28 subsets, 5mm post-filter) on 2.0x2.0x3.3mm³ voxel grids. Target volumes were estimated by SUV thresholds that accurately defined the ground-truth lesion volume. Uniform and dose-painting VMAT plans were optimized for fixed normal lung and cord objectives. Resulting plans were delivered to a cylindrical diode array at rest or driven by the same respiratory patterns on a motion platform. Errors in mean target:background ratios σ(T/B), target volumes σ(V), planned equivalent uniform target doses σ(EUD), and 3%/3mm distance-to-agreement gamma delivery passing rates σ(γ) were estimated.
Results: Relative to ground truth, 3DPET errors due to motion were σ(T/B)=-11% and σ(V)=15%. Static target doses of 60 Gy uniform or 72 Gy EUD were achieved, while motion σ(EUD)=5%. Delivery passing rates dropped from γ(static)=100% to γ(3D)=63-82%. Dose painting γ(3D) were 12% higher than uniform dose delivery for moving targets. Preliminary errors in 4DPET-defined target volumes were lower than 3DPET and not significantly different from ground truth (σ(V-4D)=-3%, p>0.18).
Conclusion: Uncertainty estimation from PET/CT imaging, RT planning, and RT delivery is feasible within an integrated respiratory motion phantom workflow. Dose-painting plans appear more robust to motion-induced delivery errors than uniform target dose plans. This motivates future investigation on patient-specific quality assurance for 4D PET/CT-guided radiotherapy, including evaluation of 4D dose-painting RT delivery.